Evidence page · writing
Brāhmī is the ancestor of nearly every script in South and Southeast Asia. Its origin is unresolved, its name comes from sources outside the brahmanical tradition entirely, and the question of where the earliest examples are found is an active and politically loaded dispute.
The word brāhmī as the name of a script appears in the Lalitavistara, a Buddhist text, in a passage where the young Buddha is examined on writing and reels off a list of sixty-four scripts. Brāhmī lipi heads the list; kharoṣṭhī is second.
And in the Jain tradition it is a person. Brāhmī is the daughter of Ṛṣabha, the first Tīrthaṅkara, and he teaches her writing — the script carries her name. Her sister Sundarī is taught arithmetic.
The two traditions that preserve the name are the two the brahmanical order spent centuries opposing. In one of them the script is named after a woman.
The modern application of the name is a nineteenth-century scholarly decision. When European epigraphists needed a label for the Aśokan script they took brāhmī from these lists. The script did not call itself that.
What the name originally referred to is not settled. The common scholarly derivation connects it to brahman or to Brahmā — a divine-origin claim of the kind traditions routinely make about writing. The Jain personification and the Buddhist list are the earliest surviving attestations of the word as a script-name; which sense came first cannot be established from them.
What can be stated without dispute: the earliest texts that preserve the name are Buddhist and Jain, and the people who used the script for its first centuries were overwhelmingly Buddhist and Jain. That is a fact about the epigraphic record, independent of where the word came from.
Both are centuries later than the script they name
| Source | Date | Distance from Aśoka |
|---|---|---|
| Aśokan inscriptions | c. 250 BCE | — |
| Lalitavistara, 64 scripts | c. 200–300 CE | ~500 years later |
| Jain Samavāyāṅga / Paṇṇavaṇā, 18 scripts | redacted c. 5th c CE | ~700 years later |
The Aśokan inscriptions do not name their own script. Neither do the Tamil-Brāhmī cave records. Nobody who used this script in its first five centuries left us a word for it.
Every name we have for the script is later than the script by half a millennium. "Brāhmī" is a label applied backwards, and "Tamil-Brāhmī" is a modern coinage on top of that.
The passage is the young Buddha's examination in writing. He names the scripts he knows, heading the list with brāhmī lipi, then kharoṣṭhī, then puṣkarasāri.
Regional scripts named in the same list include:
drāviḍa-lipi · dakṣiṇa-lipi · aṅga-lipi · vaṅga-lipi · magadha-lipi · śakāri-lipi · cīna-lipi · hūṇa-lipi · darada · khāsya
A Buddhist text of the third century CE lists a Dravidian script and a southern script by name, in the same breath as Brāhmī and Kharoṣṭhī. Whether these were genuinely distinct systems or regional hands is not recoverable from a list — but the list's compilers considered them worth naming separately.
The Samavāyāṅga and Paṇṇavaṇā give eighteen, headed by bambhī — Brāhmī. Among them:
dāmilī · javaṇāliyā · kharoṭṭhī · pukkharasāriyā · polindī · māhesarī · aṃkalivī · gaṇitalivī · gandhavvalivī
Dāmilī is the Tamil script — damiḷa, the Prakrit form of tamiḻ. An ancient source names it as its own script, not as a variety of Brāhmī.
And javaṇāliyā is yavanānī — the Greek script, the same term that appears in Kātyāyana's vārttika on Pāṇini.
So "Tamil-Brāhmī" is a modern name for something an ancient list called simply Dāmilī. The modern label embeds a claim about descent that the ancient one does not make.
South Indian megalithic pottery carries incised marks — hundreds of distinct symbols across sites in Tamil Nadu, Kerala, Karnataka and Andhra, on burial urns and domestic ware, in contexts running from roughly the second millennium BCE into the historical period.
Iravatham Mahadevan argued a continuity: that a subset of these marks corresponds to Indus signs, and that the tradition bridges the gap between the Indus system and the historical scripts of the south.
| The argument rests on | The objections |
|---|---|
| Shape correspondence between a subset of graffiti marks and Indus signs | Simple geometric shapes recur independently worldwide. Similarity is not descent without a system |
| Continuity of the marking tradition itself across the intervening period | The chronological span is wide and the dating of individual marks is often loose |
| Some marks appear alongside Tamil-Brāhmī letters on the same pots | Co-occurrence shows the two traditions overlapped, not that one produced the other |
| The graffiti are non-random and recur across sites | Non-random is not linguistic. Potters' marks, ownership marks and clan signs are all non-random and none is writing |
The honest position is that the graffiti are undeciphered marks of undetermined function, and that the continuity claim is unproven rather than refuted.
What would settle it: a full published corpus of the graffiti with secure stratigraphic dates, sign-frequency and positional analysis of the kind applied to the Indus corpus, and a test of whether the marks behave like a writing system — the same test the Indus signs themselves are subjected to.
Iravatham Mahadevan was an IAS officer. He worked on the Indus corpus and on Tamil-Brāhmī epigraphy outside any university appointment, funding his own concordance work, and produced what remains a standard reference on the Indus sign corpus.
The structural question is why the work was available to be done by someone outside the institutions. The Corpus of Indus Seals and Inscriptions was assembled largely through Finnish and Indian collaboration; the megalithic graffiti have no comparable published corpus at all.
Compare the effort. The Aśokan inscriptions were edited, translated and published within decades of decipherment. The megalithic graffiti have been known for over a century and there is still no complete catalogue with dates.
A body of evidence that would bear directly on southern antiquity has never been assembled to the standard routinely applied to northern material.
That is the same pattern the coverage page documents at Keeladi — not suppression of a result, but the absence of the work that would produce one. An argument cannot be refuted if the corpus it needs has never been published.
| Kharoṣṭhī | Brāhmī | |
|---|---|---|
| Direction | Right to left | Left to right |
| Source | Aramaic — the Achaemenid administrative script. The derivation is clear and widely accepted | Disputed |
| Range | Gandhāra and the northwest | The whole subcontinent, then Southeast Asia and Tibet |
| Fate | Extinct by about the fourth century CE | Ancestor of Devanāgarī, Tamil, Telugu, Kannada, Malayalam, Sinhala, Khmer, Thai, Lao, Burmese, Javanese, Balinese, Tibetan |
Kharoṣṭhī and Brāhmī appear in the same period and are used side by side in Aśoka's inscriptions — Kharoṣṭhī in the northwest, Brāhmī everywhere else. They run in opposite directions, which is the strongest single argument against derivation of one from the other.
The candidates for Brāhmī's origin are three, and none is settled:
1 · Semitic derivation. From Aramaic or a North Semitic script, as Kharoṣṭhī plainly is. Problem: the direction reverses, and the systematic structure differs.
2 · Indus derivation. Continuity from the Harappan signs. Problem: a gap of over a thousand years with no intervening material, and the Indus script is undeciphered so no sign-value comparison is possible.
3 · Deliberate invention. Designed, possibly for Aśoka's chancery or shortly before, by someone who knew a Semitic script and built a different system. Brāhmī's internal logic is unusually regular — it is an abugida, where each consonant carries an inherent vowel modified by systematic marks. That regularity is easier to explain by design than by drift.
The phonological ordering is the detail worth holding onto. Brāhmī's sign inventory follows the arrangement of the Sanskrit grammarians — stops by place and manner of articulation, in a sequence that reflects genuine phonetic analysis. Whoever laid it out knew the phonetics. That points to design, whatever the source of the shapes.
The conventional anchor — northern, and treated as the default clock — is Aśoka, c. 250 BCE — inscriptions that are securely dated because they name a king whose reign is fixed by cross-references to Hellenistic rulers.
Excavations in Tamil Nadu have produced dates that would push writing in the far south earlier — Kodumanal, Porunthal and Palani among them, with carbon dates from associated material running into the fifth century BCE and earlier, and Keeladi's reported eighth-century-BCE horizon.
| For an earlier southern date | Against |
|---|---|
| Published AMS dates from excavated contexts, from multiple laboratories | The dates come from associated organic material, not from the inscribed object. A dated seed in a grave dates the grave, and the sherd's relationship to it has to be argued |
| Tamil-Brāhmī has letters Aśokan Brāhmī lacks, for sounds Sanskrit does not have — ḻ, ṟ, ṉ | Additions can be later modifications of a borrowed script rather than evidence of independence |
| Graffiti marks on South Indian megalithic pottery long predate the inscriptions and may represent an earlier notational tradition | The graffiti are undeciphered and it is not established that they are writing at all |
| Northern-first is a conventional assumption that has been tested less than it has been repeated | Extraordinary claims about priority need the excavation reports published in full |
The dispute is not settled and both positions have real evidence. What is not defensible is treating northern priority as established when the southern excavations have not been fully published.
And the publication delay is itself part of the record. The coverage page here sets out the Keeladi sequence — an excavator transferred mid-project, a report submitted in 2023, a request to rework it in 2025, and an internal evaluation rejecting key findings. A dating question cannot be resolved while the excavation reports are unpublished, and the reasons they are unpublished are administrative rather than scientific.
What would settle it: direct dating of inscribed material, full publication of the stratigraphic sections, and independent specialists able to argue with the evidence rather than about the process.
Aśoka's edicts are Buddhist in sponsorship and Prakrit in language. The earliest large body of Brāhmī writing is a Buddhist king addressing his subjects in the vernacular.
The Tamil-Brāhmī cave inscriptions are overwhelmingly Jain and Buddhist — records of donations to mendicants, cut at the stone beds where ascetics lived.
Early donative inscriptions at Sanchi, Bharhut and Amaravati name their donors, and the donors are merchants, artisans, householders, women and monks — ordinary named people recording gifts, in a language they spoke.
Sanskrit does not become the normal language of inscription until several centuries later. The first substantial Sanskrit royal inscriptions are of the second century CE and after.
Writing in South Asia begins in Prakrit, under Buddhist and Jain sponsorship, recording gifts by named ordinary people. Sanskrit arrives at the epigraphic party late.
Which bears on the tradition that treats its texts as too sacred to write down. The Vedic corpus was transmitted orally by design and by insistence. The script was in use for centuries around it, doing other work for other people.